3 ms·
Could it be a statistical fluke? I'm very skeptical about reading too much into A/B results without variance/confidence interval estimates.
by arturadib 16y ago
Could it be a statistical fluke? I'm very skeptical about reading too much into A/B results without variance/confidence interval estimates.
- paraschopra 16y agoIt wasn't a fluke -- variation was statistically significant better than control (we never publish results otherwise). Clients are never happy to see their actual conversion rates and variance, etc. published in the real world, that's why I choose not to write about those figures.